INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Synchronized oscillation of membrane potentials in assemblies of neurons has been widely observed in the mammalian sensory and limbic systems (Gray and Singer 1989; Vanderwolf 1969). Network oscillations have also been observed in the olfactory systems from vertebrates (Adrian 1950; Freeman and Skarda 1985) to invertebrates (Gelperin and Tank 1990; Laurent and Davidowitz 1994). The physiological functions of these oscillations for information processing in these neural systems have not been clearly determined, but outstanding hypotheses about the function have been presented based on the experimental data in the invertebrate olfactory system (MacLeod et al. 1998; Stopfer et al. 1997; Wehr and Laurent 1996 for insect olfactory system; Delaney et al. 1994; Gervais et al. 1996; Teyke and Gelperin 1999 for molluscan olfactory system), by making the most of their advantages for the physiological experiments. For example, the isolated molluscan whole brain remains alive and it responds to an odor stimulation to the olfactory receptors, i.e., we can easily study odor-induced changes in the dynamics of membrane potentials of the neurons in the isolated molluscan olfactory CNS.

An interesting phenomenon in the network oscillations in some mammalian sensory and limbic systems is the co-existence of several types of oscillations with different frequencies in the same neural network. For example, at least three network oscillations with different frequency ranges have been observed in the mammalian hippocampus, as follows: the  wave (5-10 Hz); the  wave (40 Hz); and the sharp waves (150-200 Hz), which occur sporadically. The hippocampal  and  waves are observed during exploratory behaviors and the paradoxical phase of sleep in rats (Bragin et al. 1995; Vanderwolf 1969) and the sharp waves are observed during slow-wave sleep and awake immobility (Buzsáki 1986). The mammalian olfactory bulb, the primary olfactory structure of vertebrates, also exhibits two oscillations with different frequency ranges: one is the  wave which is induced by odor inputs and the other is a wave at several hertz which is phase-locked with respiration (Freeman and Skarda 1985). In the turtle olfactory bulb, optical imaging using a voltage-sensitive dye has revealed that three spatio-temporally different oscillations are evoked by odor inputs (Lam et al. 2000). However, in the olfactory system of invertebrates, there have been no reports on the existence of different types of oscillations in the same network system, although single-band network oscillations have been reported in the antennal lobe (Laurent and Davidowitz 1994) and mushroom body (Laurent and Naraghi 1994) in insect olfactory systems, and in the tentacle ganglion (TG) (Ito et al. 2001) and the procerebrum (PC) (Gelperin and Tank 1990) in the molluscan olfactory systems.

Taking into account some of the experimental advantages in invertebrate nervous systems, if we find the different types of network oscillations in these invertebrate olfactory systems, the systems will be, without doubt, very useful model systems to clarify differences in the neuronal dynamics and function of the two or more types of oscillations in the mammalian neural networks such as the olfactory bulb and hippocampus. In the present study, we demonstrate that the TG has two classes of network oscillations with different frequency ranges: one has a low frequency (ca. 0.5 Hz) and the other has a high frequency (ca. 1.5 Hz); the switching between these two network oscillations is regulated by serotonin. In addition, we report that odor inputs selectively modulate the low-frequency oscillation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The slugs, Limax marginatus, were anesthetized with Mg²⁺ buffer (57.6 mM MgCl₂, 5.0 mM glucose, and 5.0 mM HEPES, adjusted to pH 7.6) injected into the body cavity, and then the complex of olfactory receptors (OR) and the TG, which are located at the tip of the tentacle, was removed from the slugs (Fig. 1A). The isolated OR-TG complex was transferred to the recording chamber, which was filled with the physiological saline that contained (in mM) 70.0 NaCl, 2.0 KCl, 4.7 MgCl₂, 4.9 CaCl₂, 5.0 glucose, and 5.0 HEPES/Na (adjusted to pH 7.6). The TG network oscillations were recorded extracellularly as the oscillation of local field potentials (LFPs), using a saline-filled glass electrode whose tip diameter was 20-50 µm. The LFP was recorded from the neuropil region of the TG (Chase and Kamil 1983; Ito et al. 2000). In Figs. 1 and 2, serotonin (Sigma, St. Louis, MO) was applied to the OR-TG complex using a conventional perfusion system (see Fig. 1Aa). On the other hand, in Fig. 3, the physiological saline in the recording chamber was drained, and serotonin (100 µM) was locally applied to the TG and deodorized or odorized air (7 ml/min) was continuously applied to the ORs (see Fig. 1Ab). This system enabled us to apply an airborne odorant to the ORs. We used garlic as an odorant. The electrophysiological data were analyzed by Igor Pro (Wavemetrics), to calculate the autocorrelation, cross-correlation, and power spectrum. The frequency of the TG oscillation was determined as the frequency at the peak of the power spectrum.

View larger version (25K): [in this window] [in a new window]   Fig. 1. Two classes of network oscillations in the tentacle ganglion (TG): fast oscillation (FO) and slow oscillation (SO). A: the recording chambers in this experiment. Aa: a conventional recording chamber (used in Figs. 1 and 2). Ab: a recording chamber for odor-air stimulation (used in Fig. 3). In Ab, the olfactory receptors were exposed to air to apply the odorized air to them, while the TG component was filled with saline or serotonin (100 µM) via a glass pipette whose tip was about 100 µm in diameter. Odor information, received at the olfactory receptors (OR), is transmitted to the primary olfactory center of the TG and transmitted to the olfactory nerve (ON). B: modulation of the local field potential (LFP) oscillation in the TG by serotonin. Ba: representative traces observed in most experimental cases (n = 12/15). The TG exhibited a spontaneous 1.5-Hz network oscillation (top trace), and application of serotonin to the TG first inhibited the oscillation (middle trace), and subsequently induced a slower 0.5-Hz network oscillation (bottom trace). Scale bars are 2 s and 5 µV. Bb: the power spectrum of the LFP traces in Ba. Solid, fine dotted, and rough dotted lines are the power spectra of the upper, middle, and lower traces in Ba, respectively. Bc: the pooled data of the oscillatory frequency at the peak of the power spectrum (n = 12). C: co-existence of 2 network oscillations in serotonin. Ca: representative traces: this co-existence was occasionally observed when the application of serotonin did not inhibit the spontaneous oscillation with a 1.5-Hz frequency (n = 3/15). Raw (top) data were recorded using analog band-pass filter at 0.08-30 Hz. When the 1- to 3-Hz frequency component was subtracted by a digital notch filter, a typical serotonin-induced oscillation as shown in the lower trace in Ba was revealed (Ca; 1-3 Hz cut), while a digital high-pass filter (0-1 Hz cut) revealed a spontaneous, fast oscillation as shown in the upper trace in Ba (Ca; 0-1 Hz cut). Scale bars are 2 s and 10 µV. Cb: the auto-correlogram of the trace (Raw trace) indicating the co-existence of the 2 oscillatory components in Ca. Cc: the power spectrum of the co-existence trace in Ca (Raw trace). This spectral profile indicates the 2 peaks corresponding to the fast (the peak at 1.7 Hz; indicated by arrowhead) and the slow (the peak at 0.6 Hz; indicated by arrow) oscillations.

View larger version (27K): [in this window] [in a new window]   Fig. 2. The SO has a higher spatial coherency of oscillation than the FO. A: simultaneous recording of the LFP from 2 remote sites in the TG. The LFP events of the FO were weakly synchronous with each other at the 2 recording sites, while the LFP events of the SO were strongly synchronized. Scale bars are 2 s and 2 µV (FO) or 4 µV (SO). B: the cross-correlograms between the 2 recording sites from the traces in A. The peak value for the ordinate of the cross-correlogram indicates the synchronicity of the oscillation at the 2 remote sites in the TG, representing the spatial coherence in the TG. The peak value of the cross-correlogram in the SO is higher than that in the FO. C: the pooled data on the peak value of the cross-correlogram (n = 8).

View larger version (24K): [in this window] [in a new window]   Fig. 3. The SO is selectively modulated by stimulation with an airborne odorant. A: modulation of the FO and SO, in response to the garlic odor. In this experiment, a special chamber for odor application was used (see METHODS and Fig. 1A). The garlic odor was applied to the olfactory receptors during the period (20 s) indicated by the underlines. The oscillation amplitude of the SO was decreased in response to the odor, while the FO was hardly changed. Scale bars are 3 s and 5 µV. B: the change induced by the odor in the power spectra of the FO and SO in Fig. 3A. Solid or dotted line indicates the power spectrum during the preodor period or during the odor-stimulus period, respectively.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Limax and related terrestrial mollusks, chemical odor signals are first received at the olfactory receptors (ORs). Then, as revealed by morphological investigation, the odor signals are transmitted into two pathways: one is the pathway of ORs  TG  PC, and the other is the direct pathway from the ORs to the PC (Chase and Kamil 1983; Chase and Tolloczko 1986, 1993; Ito et al. 2000). The relative contribution of these two pathways is controversial: anterograde labeling of the epithelial pad of the ORs with horseradish peroxidase (HRP) reveals only 15% staining of all vesicle-filled processes in the TG glomeruli of Achatina fulica (Chase and Tolloczko 1986). In contrast, retrograde backfilling of the tentacle nerve, an axonic relay unit connecting the OR-TG complex structure to the PC, stained neurons of which about 88% were TG interneurons and 12% were olfactory receptor neurons in Limax (Ito et al. 2000). The former report focuses on the synaptic pathway from the ORs to the TG, and the latter one focuses on the synaptic pathway from the TG to the PC. Based on these reports, it is at least likely that output signals from the OR-TG structure to the PC are largely derived from the TG interneurons in Limax, which clearly indicates the importance of the ORs  TG  PC pathway in the odor-information processing. Then, we regarded the TG as the primary olfactory center of Limax, and the PC as the secondary (following) olfactory center, although the shortcut pathway from the ORs to the PC is not relayed by the TG. In the present study, we focus on the TG, the primary olfactory structure of Limax.

The TG of Limax also shows a synchronized oscillatory network similar to the olfactory systems of the other species (Ito et al. 2001). A recent study using anti-serotonin antibodies indicated that there are no serotonergic somata in the TG, but numerous serotonin-containing fibers are observed in the neuropil region of the TG (H. Suzuki, personal communications). We first applied serotonin (100 µM) to the TG and recorded the changes in the LFP oscillation in the TG (Fig. 1, B and C). The TG exhibits a spontaneous fast-frequency network oscillation at about 1.5 Hz (Fig. 1Ba, top trace; and see the solid line of the power spectrum in Fig. 1Bb). In most cases (n = 12/15), application of serotonin first inhibited the spontaneous high-frequency oscillation (Fig. 1Ba, middle trace; and fine dotted line in Fig. 1Bb) and subsequently induced a slow-frequency and larger amplitude network oscillation at about 0.5 Hz (Fig. 1Ba, bottom trace; and rough dotted line in Fig. 1Bb). The averaged frequency is 1.69 ± 0.18 Hz in saline, and 0.47 ± 0.05 Hz in serotonin (Fig. 1Bc; mean ± SE; n = 12). In some cases (n = 3/15), application of serotonin did not inhibit the spontaneous 1.5-Hz oscillation, which resulted in co-existence of two network oscillations (1.5- and 0.5-Hz oscillation) in the TG (Fig. 1Ca; Raw). The auto-correlogram (Fig. 1Cb) and power spectrum (Fig. 1Cc) of the raw LFP trace in Fig. 1Ca also indicate that different oscillatory modes with different frequency ranges co-exist in the TG. These results indicate that serotonin induces a new type of network oscillation with a low frequency (0.5 Hz) in the TG, rather than modulates the existing 1.5-Hz oscillation by decreasing its frequency. Thus the TG is an oscillatory network that can express two types of network oscillations. We named the spontaneous 1.5-Hz oscillation the fast oscillation (FO), and named the serotonin-induced 0.5-Hz oscillation the slow oscillation (SO). The transition between FO and SO in the TG is determined by the extracellular level of serotonin.

Next, we investigated the difference between the two oscillations, FO and SO, with respect to their spatial coherence (Fig. 2). For this purpose, we recorded the LFP oscillations at two sites distant from each other in the TG neuropil region (n = 8). Two LFP recording electrodes were laterally located at the opposite sides of the TG (see Fig. 1Aa): the inter-electrode distance was approximately 200 µm. The LFP events of the FO were weakly synchronized at the two remote sites in the TG: these LFP events were synchronized in some cases, but not synchronized in the other cases (Fig. 2A; Fast Oscillation). On the other hand, the LFP events of the SO were always strongly coupled (Fig. 2A; Slow Oscillation). The peak value of the cross-correlogram of the LFP oscillations at two remote sites provides an index of the spatial coherence between the two sites. They revealed that the LFP events during the SO occurred more coincidentally than those during the FO (Fig. 2, B and C). The peak values of the cross-correlogram were 0.39 ± 0.02 for the FO and 0.73 ± 0.06 for the SO (mean ± SE) and the difference is statistically significant (Fig. 2C; P = 0.0002 with a paired t-test). These results indicate that the FO and SO are different in their spatial coherence in the TG as well as the oscillation frequency: the FO is a spatially noncoherent mode and the SO is a coherent mode.

Odor information is first received at the ORs and subsequently transmitted to the TG, Limax's primary olfactory center. Finally, we investigated the modulation of the oscillatory activity in the TG in response to olfactory stimulation. When we applied an odorant, garlic, to the ORs, the dynamics of the FO were not changed (Fig. 3A; FO). However, we found that the SO was remarkably modulated with its amplitude in response to the garlic odor, i.e., the odor input decreased the oscillation amplitude of the SO (Fig. 3A; SO). These findings were confirmed by the change of the power spectrum of the TG oscillation in response to the odor input, calculated from the LFP traces in Fig. 3A (Fig. 3B). The quantitative analysis (n = 3) revealed that the power of the SO (0.25-0.75 Hz) was reduced in response to the odor (36.3 ± 6.4%), while the power of the FO (1.5-2.0 Hz) is hardly changed in response to the odor (107.7% ± 29.2%). Thus these results indicate that the SO is selectively modulated by an odor stimulus in the TG oscillatory dynamics.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the present study, we first found that the TG, the molluscan primary olfactory center, has two classes of network oscillations, the FO (1.5 Hz) and the SO (0.5 Hz) (Figs. 1 and 2). The switching between the FO and SO is regulated by the extracellular level of serotonin. Serotonin also induces a more spatially coherent oscillatory synchronization throughout the TG (Fig. 2). In addition, we clarified that the SO is selectively modulated by odor inputs (Fig. 3). Thus in the mode of SO, olfactory information is more strongly influenced by the temporal architecture of the synchronized oscillation than in the mode of FO. This is the first report in invertebrate olfactory systems, including insects and mollusks, to identify the existence of more than one class of network oscillation and to clarify the difference with respect to the odor-induced dynamics of these oscillations.

From our new findings in the present paper, two future directions would emerge: The first question is how olfactory information processed in the TG is outputted to the PC, the secondary (following) olfactory center of Limax, in the FO mode and in the SO mode. The second question is in what kind of situation serotonin is released and the FO mode is changed to the SO mode. Regarding the first question, we have no direct evidence at present for the decoding of the TG activity. However, it is widely accepted that, due to biophysical constraints, "oscillatory" activity locks the timing of firing in a single neuron at the peak of the oscillating membrane potential, while "synchronized" activity brings about coherent discharges among a population of neurons. Taking these into account, during odor processing in the noncoherent and odor-insensitive FO mode, the firing in each TG neuron evoked by odor-input is still temporally constrained by the FO oscillatory activity, because the FO is odor-insensitive and still exists during odor presentation. In addition, since the FO is a globally noncoherent but locally coherent mode, locally synchronized discharges in the TG would be outputted to the PC in response to odor input. On the other hand, in the globally coherent and odor-sensitive SO mode, the constraint by oscillatory and synchronized activity disappears in response to odor input, and hence a disordered pattern of discharges would be delivered to the PC. Thus different properties of the FO and the SO with respect to coherency and odor-responsiveness could give a different modulation in the PC activity via different spatio-temporal firing patterns in the TG. This implies different physiological functions of the FO and the SO in odor coding. We recognize that a combined approach of electrophysiological technique of single neuron recording and computational network modeling of the TG and the PC is adequate to directly test the idea described above.

Regarding the second question, in some gastropod mollusks such as Aplysia (Abrams 1985; Hawkins et al. 1993) and Helix (Balaban et al. 1987), serotonin has an important role in memory acquisition. In the classical conditioning of the gill-withdrawal reflex in Aplysia, serotonin is liberated at the targeted synapses from the pathway of unconditioned stimulus such as a tail shock (Hawkins and Schacher 1989). It is plausible that a serotonin-induced SO is observed in the process of odor-memory acquisition in Limax, via the unconditioned stimulus pathway. It will be interesting to examine whether the neural switching between the FO and the SO contributes to the functional switching between the odor-recognition process (when only the conditioned stimulus is coming) and odor memory-acquisition process (when the conditioned stimulus and unconditioned stimulus are coming concomitantly), respectively. It is necessary to identify the origin of the serotonergic pathways to the TG to test this hypothesis.

The olfactory systems in invertebrates have some experimental advantages: when we use the molluscan olfactory systems as a model system for mammalian olfaction, we can study odor-induced neuronal dynamics in the olfactory system in vitro (Delaney et al. 1994; Gelperin and Tank 1990), which enables us to more easily record odor-induced changes of the membrane potentials in a single neuron of the olfactory system, than in in vivo recording. As preliminary data, we recently succeeded in recording the membrane potential of a single neuron in the TG, in response to odors and serotonin (Inokuma et al., unpublished observations). The switching between the FO and SO by serotonin (Figs. 1 and 2) and the difference in the odor responsiveness (Fig. 3) in the TG of the mollusk would be an interesting model mechanism to elucidate the differential functions of multi-mode network oscillations in the olfactory information processing at the neuron-network level.